vol. 167, no. 6 the american naturalist june 2006

Pereskia and the Origin of the Life-Form

Erika J. Edwards* and Michael J. Donoghue†

Department of Ecology and Evolutionary Biology and the Peabody survival is conferred by a suite of anatomical and physi- Museum of Natural History, Yale University, P.O. Box 208105, ological specializations. All investigated cacti exhibit ex- New Haven, Connecticut 06511 tensive, shallow rooting systems, which allow for the rapid Submitted September 2, 2005; Accepted February 21, 2006; absorption of water from the top layer of after brief Electronically published April 28, 2006 . Most cacti possess enlarged pith and stem cortical layers, which contain large mucilaginous cells that aid in long-term water storage. Additionally, most cacti exhibit crassulacean acid metabolism (CAM) photosyn- thesis, a specialized pathway that temporally separates abstract: The cactus life-form is cited as an example of a tight relationship between organism form and function: a succulent, long- atmospheric CO2 uptake from photosynthetic light reac- lived, photosynthetic stem allows cacti to survive long periods of tions, allowing stomata to open during nighttime when while maintaining a positive water status. the transpirational water loss is reduced (Kluge and Ting (Cactaceae) comprises 17 of leafy and that are 1978). Finally, most cacti are functionally leafless and have thought to represent the original cactus condition. Recent phylo- transferred their primary photosynthetic activities to their genetic work has shown that there are two separate clades of Pereskia long-lived, specialized stem cortical tissue layer. species, which are and paraphyletic with respect to the rest of the cacti. We selected seven Pereskia species, representing both clades, The elimination of is thought to be among their and characterized their water relations by measuring a suite of phys- most important adaptations to drought: leaves are tem- iological traits in wild populations. Additionally, we estimated basic porary structures with large surface areas that allow for climate parameters from collection localities for all 17 Pereskia spe- excessive water loss to the atmosphere. Creating a long- cies. Extant Pereskia species exhibit ecological water use patterns that lived photosynthetic tissue system in the stem minimizes are very similar to those of the leafless, stem-succulent cacti. Ancestral the exposure of hydrated tissue to the atmosphere while trait reconstruction for the physiological and environmental data at the same time extending the potential lifetime carbon provides a preliminary assessment of the ecology and water relations of the earliest cacti and suggests that several key elements of the assimilation of photosynthetic cells, which allows them to cactus were established before the evolution of the be more conservative about opening their stomata. Taken cactus life-form. We interpret these ecological traits as potentially together, these traits promote a highly successful “ecolog- important drivers of evolutionary innovation in the cacti. ical water use strategy”: cacti live in environments char- acterized by extreme drought but maintain positive tissue Keywords: key innovation, Pereskia, Cactaceae, ancestral trait recon- water status by being exceptionally good at acquiring and struction, water relations, ecological niche, character evolution. storing water and simultaneously exhibiting high photo- synthetic water use efficiency (Nobel 1977, 1988; Barci- kowski and Nobel 1984; Gibson and Nobel 1986). The cactus life-form is often heralded as a striking example Pereskia (Cactaceae) consists of 17 species of leafy shrubs of adaptive evolution in . Most cacti have the ability and trees that have long been considered the best living to survive extended periods of extreme drought, which has representation of the “ancestral cactus” (fig. 1). Several allowed the lineage to radiate extensively throughout the phylogenetic studies have questioned the of ’s arid and semiarid (Gibson and Pereskia, though they are all limited by taxon sampling Nobel 1986; Nobel 1988; Anderson 2001). Their drought and/or resolution (Wallace 1995; Nyffeler 2002; Butter- worth and Wallace 2005). A new phylogenetic hypothesis * Present address: Geography Department, University of , Institute for Computational Earth System Science 3060, Santa Barbara, California of basal cactus relationships based on molecular data re- 93106; e-mail: [email protected]. solves “Pereskia” as a paraphyletic assemblage of species † E-mail: [email protected]. at the base of the cacti and confirms that their leafy Am. Nat. 2006. Vol. 167, pp. 777–793. ᭧ 2006 by The University of Chicago. and nonsucculent stems were not secondarily derived (Ed- 0003-0147/2006/16706-41282$15.00. All rights reserved. wards et al. 2005). Further, the Pereskia species united with Figure 1: Pereskia growth form and morphology. Pereskia species range from woody scramblers and shrubs to taller trees, reaching over 20 m in height in Pereskia lychnidiflora. A, The Pereskia portulacifolia. B, The Pereskia guamacho. C, Leafy stem of Pereskia sacharosa, showing the use of the areole short- system to produce leaves as well as spines. D, Succulent leaves subtending the spiny cactus areole during new stem growth in Pereskia weberiana. Origin of the Cactus Life-Form 779 cactoid, opuntioid, and lineages (referred to as the “core cacti”) have stem stomata and exhibit delayed formation, two traits that are critical in the devel- opment of the stem as a long-lived photosynthetic (Edwards et al. 2005). This substantiates the usefulness of Pereskia species for inferring early events in the transition to the cactus life-form and allows us to address several key questions regarding the evolution of the specialized cactus water use strategy. How did the cacti regulate their water use before the evolution of efficient water storage tissue systems and stem-based ? What sorts of environments did they live in, and what levels of drought could they tolerate? Are there particular aspects of their ancestral water relations that may have promoted the evolution of the specialized cactus life-form?

Background and Objectives In spite of the long-standing interpretation of Pereskia as a model of the ancestral cactus, there has been little work done to characterize basic Pereskia ecology and physiology. Historically, Pereskia has been described as inhabiting a range of dry tropical forest areas but not subject to or surviving severe water stress. They are generally considered to be drought-, nonsucculent woody plants, with primarily C3 photosynthesis and weak CAM-cycling abil- ities (Rayder and Ting 1981; Gibson and Nobel 1986; Leuenberger 1986; Nobel and Hartsock 1986, 1987; Mau- seth and Landrum 1997; Mauseth 1999; Martin and Wal- Figure 2: Basic parameters of water relations. Water flux through the soil-plant-atmosphere continuum is generally modeled after Ohm’s lace 2000). CAM cycling refers to a variant of CAM pho- law, where the driving force behind water movement is the difference tosynthesis wherein the plant opens stomata by day and between soil and atmospheric water potentials (DW). The rate of water closes them at night but uses the CAM metabolic pathway flux is a function of DW and the hydraulic conductivity of the pathway.

Plant hydraulic conductivity (Kh) is often compartmentalized into , to reassimilate internally respired CO2 (Kluge and Ting stem, (K ), and leaf boundary layer components, each of which 1978). Investigations of photosynthetic pathway variation lamina may vary independently. To compare Kh of xylem (the primary plant have primarily utilized greenhouse-grown plants, and field water-conducting tissue) across plant species, researchers usually nor- studies of wild Pereskia populations have been limited (but malize stem Kh by either the amount of conductive tissue measured see Diaz 1984; Diaz and Medina 1984; Luttge et al. 1989). (sapwood cross-sectional area, Ksp) or the amount of leaf area that the Even less is known regarding how Pereskia species reg- measured conductive tissue is supplying (one-sided leaf surface area, KL). K is an intrinsic value of the , largely governed by vessel length, ulate their water use. General characteristics of the plant sp diameter, and density, while KL reflects both Ksp and Huber value (HV), water transport pathway and the processes involved in which is the sapwood/leaf area ratio. A high HV means that plants are regulating plant water loss are illustrated and explained in investing in large amounts of wood per leaf, which is a relatively expensive figure 2 (see table 1 for terminology abbreviations and (in terms of both carbon and energy costs) allocation pattern. definitions). Plants living in extremely water-limited en- vironments exhibit several water use strategies that differ potentials -round because of a high leaf-specific xylem in phenology, leaf life span, rooting depth, minimum tol- conductivity (KL) conferred by a large wood-to-leaf carbon erated tissue water deficits (pminimum tissue water po- allocation pattern (HV). Stomatal behavior of P. guamacho tential, Wmin), maximum rates, and growth was conservative and afforded a high photosynthetic water form. use efficiency. Additionally, P. guamacho exhibited complex Edwards and Diaz (2006) recently investigated the eco- and unpredictable leaf phenological patterns; different physiology of Pereskia guamacho in northwestern Vene- populations were asynchronous with one another as well zuela and found that the water relations of P. guamacho as with their respective plant communities. Rather than are strikingly different from those of co-occurring woody, shedding its leaves in response to drought, one population leafy plants. Pereskia guamacho maintained high leaf water of P. guamacho performed drought-induced CAM pho- 780 The American Naturalist

Table 1: List of abbreviations used in text Abbreviation Definition Unit Ϫ1 Ϫ1 Ϫ1 Ksp Sapwood-specific xylem hydraulic conductivity kg m s MPa Ϫ4 Ϫ1 Ϫ1 Ϫ1 KL Leaf-specific xylem hydraulic conductivity 10 kg m s MPa HV Huber value 10Ϫ4 m2 mϪ2 SPI Stomatal pore index 10Ϫ2 mm2 mmϪ2 SLA Specific leaf area m2 kgϪ1

Wmin Minimum bulk leaf water potential MPa d13C Leaf carbon isotope discrimination ratio ‰ tosynthesis, while a second retarded leaf expansion and water use. For each of the seven species, we measured Ϫ1 Ϫ1 Ϫ1 kept stomata closed day and night. Edwards and Diaz sapwood (Ksp,kgm s MPa ) and leaf-specific xylem Ϫ1 Ϫ1 Ϫ1 (2006) speculated that P. guamacho is not strictly drought hydraulic conductivity (KL,kgm s MPa ), Huber deciduous and may instead perform drought-induced value (HV), minimum leaf water potential (Wmin, MPa), Ϫ2 CAM photosynthesis as a means of lengthening leaf life and daily patterns of stomatal conductance (gs, mmol m span. They concluded that in many ways, the water use sϪ1). All field measurements were made in natural pop- strategy of P. guamacho is similar to that of the leafless, ulations during the growing (rainy) season on healthy ma- stem-succulent cacti: it too maintains a positive tissue wa- ture plants growing in full sun, using methods identical ter status while surviving in water-limited environments, to those described by Edwards and Diaz (2006). In Pe- it exhibits water storage (in leaves), and its stomatal be- reskia, it proved difficult to measure Klamina directly because havior is conservative and tightly regulates water loss. To of succulence and the lack of a in many species. put it simply, P. guamacho may not look like a cactus, but However, Sack et al. (2003) demonstrated a highly sig- it behaves like one. nificant correlation between stomatal pore index (SPI; If P. guamacho is representative of all Pereskia species, guard celllength2 # stomatal density [mm2 mmϪ2]) and this suggests that the morphological and anatomical spe- leaf laminar conductance (Klamina) across a diverse collec- cializations exhibited by the core cacti are not directly tion of species, so we instead calculated the SPI of three responsible for their ecological water use strategy per se; field-grown leaves from multiple individuals from each rather, key elements of this strategy were more or less species, and we here use SPI as a proxy for Klamina. established in the Cactaceae before the evolution of stem Stable carbon isotope ratios of leaf tissue are routinely succulence and the evolutionary loss of functional leaves. used as a time-integrated measure of photosynthetic water Here we characterize the physiological ecology of other use efficiency for C3 plants, with lower levels of 13C dis- Pereskia species, focusing especially on traits pertaining to crimination indicative of higher water use efficiency (Far- water use, and we use this information to infer the ecology quhar et al. 1982). They are also used to differentiate be- and water relations of the first cacti. We extend the sam- tween photosynthetic pathways because the used pling of Pereskia ecophysiology to include six additional in the first carboxylation step of atmospheric CO2 differ species, representing both major Pereskia clades. In order strongly in their discrimination of 13C (Kluge and Ting to more accurately characterize the climate regimes of ex- 1978). Multiple leaves from each individual (n p 3 in- tant Pereskia, we gathered climate data from specimen col- dividuals for each species) were dried, bulk ground with lection localities for all Pereskia species. We then recon- mortar and pestle, and subsampled. Using a Finnigan MAT structed both ecophysiological and climatic variables at delta E isotope ratio mass spectrometer, 13C/12C ratios were two specific nodes in the basal cactus phylogeny to help determined on CO2 collected from the samples after com- infer the ecological water use strategy of ancestral Pereskia, bustion. Numbers here are expressed relative to the PDB and we used these reconstructions to construct a prelim- standard using the equation inary hypothesis of the ecological and physiological con- ditions that preceded the evolution of the typical cactus 13C/ 12 C life-form. d13C p 1,000 sample Ϫ 1. ()13C/ 12 C standard

Methods We also estimated specific leaf area (SLA, m2 kgϪ1), cal- culating bulk leaf surface area and dry weight of multiple Ecophysiological and Anatomical Data leaves for eight individuals of each species. SLA is a func- To characterize the water relations of Pereskia species, we tion of leaf water content, leaf thickness, and leaf density focused on a suite of traits relating to plant hydraulics and and has been shown to correlate negatively with leaf life Origin of the Cactus Life-Form 781 span and nutrient use efficiency and positively with max- ter et al. (1997). lengths, intraspecific trait varia- imum photosynthetic rate (Reich et al. 1992; Ackerly and tion, and assumptions regarding models of character evo- Reich 1999). It is often used as a proxy for carbon in- lution can all have significant consequences in estimating vestment per leaf and as a good predictor of ecological ancestral values (Felsenstein 1985; Donoghue and Ackerly growth strategy (Westoby et al. 2002). 1996; Martins and Hansen 1997; Cunningham et al. 1998; Butler and King 2004). To explore the sensitivity of our reconstructions to each of these variables, we generated Climate Data sets of 500 trees with randomized branch lengths (using To create a more accurate picture of current Pereskia ecology, the “generate trees” function in COMPARE 4.6) and ran we collated all specimen collection location information in multiple analyses across all trees, using different models Leuenberger (1986) and from the collections of the Missouri of phenotypic trait evolution. In all analyses, we included (St. Louis, MO) that are available online. standard errors of our trait means as estimates of intra- We then translated the descriptive location data into specific trait variation. We employed two primary evolu- latitude/longitude coordinates using the free Web-based tionary models: a Brownian motion (BM) model of trait tool BioGeomancer (http://www.biogeomancer.org). We evolution, which corresponds to Schluter et al.’s (1997) used DIVA-GIS (http://diva-gis.org/) to map these points maximum likelihood method and assumes that traits and extract climate information for each species distribu- evolve by drift; and a simple Ornstein-Uhlenbeck (OU) tion. DIVA-GIS uses the WorldClim global climate data set, model with one evolutionary optimum, which may be available at http://www.worldclim.org. We calculated mean more appropriate for traits evolving under stabilizing se- and standard error of each climate variable for each species lection (Hansen 1997). Butler and King (2004) present a and used these as tip values in our phylogenetic climate compelling argument for exploring more complex OU reconstructions. models with multiple evolutionary optima; however, our limited taxon sampling (n p 3–9 taxa per tree) prevents us from estimating such parameter-rich models with any Ancestral Trait and Climate Reconstruction confidence. We did explore effects of selection strength on Figure 3 depicts the basal cactus phylogeny described by ancestral reconstruction, however, by employing four dif- Edwards et al. (2005). The current lack of trait data from ferent values for a (0.5, 1.0, 5.0, and 10.0). For each of cactus outgroups as well as appropriate members of the the Pereskia clades (separately), we ran BM and OU models core cacti imposes important limitations on how confi- across 500 trees with randomized branch lengths to re- dently we can infer the character states of our particular construct seven physiological traits and eight climate var- physiological traits for the basal nodes in Cactaceae. In iables for nodes A and B (fig. 3). For the physiological some instances, traits that are highly relevant to Pereskia traits, we used trees that consisted of only the seven focal water use are irrelevant to or even impossible to measure Pereskia species (two trees, one of four and one of three in the core cacti; leaf-related traits, such as SLA or Klamina, species), and for the climate reconstructions, we used trees for example, cannot be measured in a plant with no func- that included all 17 Pereskia species (two trees, one of eight tional leaves. For this reason we chose to reconstruct the and one of nine species). basal nodes of the two Pereskia lineages separately, and we Using environmental parameters of extant species to use these values to infer the ecological setting and water infer ancestral climates is a relatively novel endeavor (Gra- relations of ancestral Pereskia (fig. 3, nodes A and B). Our ham et al. 2004; Hardy and Linder 2005), and the con- fundamental assumption is that a general ecological water ceptual basis of this approach has not yet been fully de- use strategy that is shared by both Pereskia clades and the veloped. For example, most environmental parameters are core cacti is likely to be ancestral for Cactaceae. This would continuous variables, such as mean annual , present a working hypothesis, to be tested as we gather and a given species distribution will encompass a range more physiological data from portulacaceous outgroups of values for this variable. Graham et al. (2004) and Hardy and appropriate core cacti, and as basal and and Linder (2005) reconstructed minimum and maximum phylogenetic relationships become better values for species independently and used these to delimit resolved. “ancestral niche envelopes.” This may be overly conser- We used COMPARE 4.6 (Martins 2004) to explore dif- vative because the density curves of sampled environ- ferent methods of ancestral trait reconstruction for con- mental parameters should be similar to those of other tinuous characters. COMPARE employs a generalized least continuous organismal traits (e.g., with appropriate sam- squares model of ancestral state estimation (Martins and ple size, they are normally distributed). Such is the case Hansen 1997), which in its simplest form is analogous to for many of our climate variable distributions (see fig. 4). the maximum likelihood reconstruction method of Schlu- Because we are most interested in the probable climates 782 The American Naturalist

Figure 3: Basal cactus phylogeny, adapted from Edwards et al. (2005). Phylogram generated from a maximum likelihood search using a concatenated five-gene-region data set of 6,450 characters. Pereskia is paraphyletic, with nine Pereskia species united with the core cacti in the “caulocactus” clade. Species chosen for ecophysiological field studies are shown in boldface capital letters. We reconstructed ancestral values of nodes A and B for seven physiological traits, using measured values from the seven focal Pereskia species. For reconstructed climate estimates of nodes A and B, we used data from all 17 Pereskia species. at nodes A and B, as opposed to the full range of potential exhibit the same general water use pattern described for climates, we have used means and standard errors for the Pereskia guamacho by Edwards and Diaz (2006). Values of extant species in our reconstructions. KL and HV are high, allowing for an efficient water supply system to transpiring leaves (this is taken to the extreme in Pereskia portulacifolia, with K and HV values among Results L the highest reported for any woody broad-leaved plant).

Ecological Physiology of Extant Pereskia Species The value for SPI, our proxy for Klamina, is among the lowest recorded in the literature. A coupling of high K and low Means and standard errors of seven ecophysiological traits L K implies that whole-plant water use is being regulated for the focal Pereskia species are reported in table 2, to- lamina gether with values culled from the literature to provide primarily at the leaf level. context for the Pereskia values. When possible, studies Pereskia minimum leaf water potentials are remarkably from tropical dry forest plant communities were used for high for woody plants of semiarid tropical environments. 13 comparison; unfortunately, this was not possible for SPI The d C values indicate very high photosynthetic water because this trait has not yet been reported from species use efficiencies, with extremely high values in Pereskia diaz- living in these systems. Also, while there are many studies romeroana and Pereskia sacharosa. In a survey of CAM that measure xylem hydraulic properties, the methods used plants that exhibit plasticity in the proportion of atmo- to do this are not standardized, making comparisons across spheric carbon fixed during the day or night, Winter and studies difficult. For this reason, the range of values we Holtum (2002) found a strong linear relationship between 13 present for Ksp, KL, and HV are from one study in a Costa tissue d C and the percentage of carbon uptake occurring Rican dry forest (Brodribb et al. 2002) whose methods at night, with d13C values of Ϫ21‰ and Ϫ22‰ corre- were the same as those used here. sponding to approximately 20% nocturnal carbon uptake. Despite large morphological differences between the This suggests that P. diaz-romeroana and P. sacharosa, like seven focal Pereskia species (fig. 1), their ecophysiological P. guamacho (Edwards and Diaz 2006), are using the CAM characteristics are generally quite similar, and all species photosynthetic pathway to some degree. Values for SLA Figure 4: Box plots of two environmental parameters of extant Pereskia species. Values inside boxes represent 75% of data points, and tails include 95% of data points; line inside box is median. The closer the median line is to the center of the box, the more normally distributed are the data. Numbers next to species names are the total number of collection localities representing that species. Species names in boldface were sampled for ecophysiological trait characterization; the selected species encompass the majority of climate variation experienced across all Pereskia species. A, Mean annual precipitation; B, mean annual temperature. Most climate regimes correspond to the tropical dry or very dry forest Holdridge life zone (Holdridge 1967). Table 2: Values of ecophysiological traits measured for seven Pereskia species and reconstructed nodes A and B p p p 13 Study site Latitude and Ksp (n 8 KL (n 8 HV (n 8 SPI SLA Wmin C Species location longitude )a branches)a branches) (n p 3 leaves) (n p 8 trees) (n p 5 trees) (n p 3 trees) 42. ע Ϫ21.39A 09. ע Ϫ.88AB 39. ע 13.80B 21. ע 3.38BC 1.18 ע 5.54B 97. ע 7.81BC 30. ע Pereskia diaz-romeroana 18.10113ЊS, 1.69AB 64.45551ЊW 31. ע Ϫ25.83B 04. ע Ϫ1.00A 1.27 ע 15.09B 26. ע 2.74C 66. ע 5.79B 1.44 ע 5.91BC 26. ע Pereskia guamacho 11.28126ЊN, 1.03B 69.69042ЊW 58. ע Ϫ26.17B 05. ע Ϫ.60C 2.07 ע 24.80A 53. ע 5.18AB 91. ע 5.62B 1.75 ע 11.16B 20. ע Pereskia marcanoi Dominican 19.0895ЊN, 2.06AB Republic 71.68403ЊW 33. ע Ϫ24.3BC 04. ע Ϫ.84AB 84. ע 11.45B 89. ע 6.02A 1.54 ע 10.82A 2.37 ע 27.62A 27. ע Pereskia portulacifolia Dominican 18.42655ЊN, 2.69ABC Republic 71.76983ЊW 58. ע Ϫ25.43B 03. ע Ϫ.71BC 2.43 ע 19.36A 30. ע 3.98ABC 56. ע 2.20B 54. ע 3.60C 14. ע Pereskia quisqueyana Dominican 18.36913ЊN, 1.76AB Republic 68.84258ЊW 28. ע Ϫ22.54AC 04. ע Ϫ.84AB 1.90 ע 13.58B 02. ע 3.45BC 63. ע 4.65B 1.44 ע 8.85BC 25. ע Pereskia sacharosa Bolivia 18.27388ЊS, 2.01AB 64.15802ЊW 72. ע Ϫ26.12B 03. ע Ϫ.73BC 72. ע 13.29B 04. ע 3.36BC 12. ע 1.96B 2.60 ע 4.56BC 97. ע Pereskia weberiana Bolivia 16.54096ЊS, 1.91AB 67.39115ЊW Reconstructed ancestral values: 000. ע Ϫ25.41 000. ע Ϫ.82 005. ע 16.16 003. ע 4.06 003. ע 5.35 110 . ע 8.89 001. ע Node A 1.76 002. ע Ϫ22.99 000. ע Ϫ0.82 000. ע 13.69 000. ע 3.40 005. ע 3.33 002. ע 7.54 000. ע Node B 1.87 Comparable values for other broad-leaved C3 angiosperms 1–3.8 .48–4.26 .54–1.7 3–20 4.8–26.8 Ϫ1.0 to Ϫ5.0 Ϫ22 to Ϫ30 Sources of comparable values Brodribb et al. Brodribb et al. Brodribb et al. Sack et al. Eamus and Sobrado 1986; Mooney et al. 2002 2002 2002 2003 Prior 2001; Eamus and 1989; Smith et Vendramini et Prior 2001; al. 1997; al. 2002 Brodribb et Eamus and al. 2002; E. Prior 2001 Edwards, unpublished data ,SE . For extant species, values with different letters are significantly different from one another (all pairs Tukey-Hamer test,P ! .05 ). For definitions of abbreviations ע Note: Values are presented asmean see table 1. a With the exception of P. weberiana, wheren p 2 branches. Origin of the Cactus Life-Form 785

ranged from 11.45 m2 kgϪ1 in P. portulacifolia to 24.8 m2 kgϪ1 in Pereskia marcanoi, which fall in the middle to higher end of values for tropical dry forest trees (Eamus and Prior 2001) and are considerably higher than values reported for fully succulent plants (Vendramini et al. 2002). In addition to these traits, we monitored daily patterns

of stomatal conductance (gs) for the seven species, which provided further evidence for conservative water use in

Pereskia (fig. 5). Maximum gs recorded varied between 90 and 183 mmol mϪ2 sϪ1 and usually peaked in midmorning, with stomata often closing for much of the afternoon. In several species (P. guamacho, P. marcanoi, P. sacharosa), stomata remained closed for entire days during their pri- mary growing season, while co-occurring trees were tran- spiring freely (data not shown). Pereskia stomatal behavior is also remarkably sensitive to rainfall events, as witnessed in P. marcanoi and P. diaz-romeroana (fig. 5); this suggests that these species have extensive, shallow root systems sim- ilar to those of the leafless, stem-succulent cacti. A small degree of stem stomatal conductance was recorded in Pe- reskia weberiana, a member of the caulocactus clade char- acterized by stem stomata and delayed bark formation. Conductance values were so low in comparison to Pereskia leaves, however, that it is highly unlikely that significant carbon assimilation occurs in the stem. This is confirmed by the carbon isotope data; d13C values of leaf and stem cortical tissues within an individual plant were identical (data not shown), suggesting one site of carbon assimi- lation. These results support the Pereskia ecological water use strategy hypothesized by Edwards and Diaz (2006) based on P. guamacho. The combination of water relations traits exhibited by Pereskia is unusual for woody, leafy plants. In particular, comparative studies have recently found sig- nificant coordination between plant hydraulic and pho-

tosynthetic capacity, such that a high KL supports a high

Klamina, high gs, and high photosynthetic rates (Brodribb and Feild 2000; Brodribb et al. 2002, 2003; Meinzer 2003; Sperry et al. 2003; Bucci et al. 2004; Santiago et al. 2004; Sack and Tyree 2005). While we do not have direct mea- sures of photosynthesis, SLA values suggest that maximum net photosynthesis will be moderate in comparison with

diaz-romeroana was similar to that of P. marcanoi, with a more substantial Figure 5: Patterns of stomatal conductance in Pereskia. A, Pereskia mar- stomatal response to rainfall events than in its woody co-occurring coun- canoi, with co-occurring Guaiacum officinale as a control. Stomata of P. terparts, here represented by Capparis speciosa. C, Pereskia weberiana, marcanoi minimally opened for only a brief period in the morning until showing again a brief period of maximal gs in the early morning, followed there was a substantial rainstorm, after which stomata remained open by stomatal closure for much of the day. Stems of P. weberiana also for most of the day. Maximum gs was quite low in comparison to that exhibited some stomatal conductance, though substantially less than the of co-occurring trees. Breaks in the curve represent periods where no leaves. Stem gs peaked just before dawn, indicating CAM-like patterns measurements were made. B, Observed stomatal behavior in Pereskia of stomatal opening. 786 The American Naturalist other broad-leaved trees of tropical dry regions. Pereskia small size of our trees and the similarity of tip values. plants are hydraulically well built, and they could con- Node B was consistently estimated with slightly lower KL, ceivably support high leaf transpiration rates, but stomatal SPI, HV, and SLA and higher Ksp and photosynthetic water behavior is conservative, and SPI (our proxy for Klamina and use efficiency than node A (table 2). Node B habitat also a good measure of gmax) is low. In other words, the leaves may have had lower mean annual precipitation with more of Pereskia are generally “oversupplied” with water. seasonal rainfall and cooler, more highly seasonal tem- It is currently not clear why Pereskia species do not take perature patterns (table 3). advantage of their high KL and keep stomata open a greater When viewed within a broad ecological context, how- proportion of the time; neither do we know why Pereskia ever, the differences in reconstructed values for nodes A species apparently avoid the development of lower Wmin and B are relatively slight. In general, our results imply in their leaves. In a sense, however, the unusual leaf-stem that ancestral Pereskia inhabited tropical, semiarid to sub- relationship in Pereskia bears a functional resemblance to humid environments with discernible “wet” and “dry” - that of the inner and outer cortical tissue of the core cacti: sons and relatively low mean annual rainfall, but they as a cactus stem dehydrates, the inner cortex behaves as nevertheless maintained very high minimum water po- a water reservoir for the outer, photosynthetic cortical tentials in their photosynthetic tissues. They accomplished layer, thus maintaining a steady and reliable water supply this in two ways: first, by allocating large amounts of water- to transpiring tissues (Barcikowski and Nobel 1984; Lerdau conducting tissue (wood) per given amount of leaf area, et al. 1992). While the xylem in Pereskia is not explicitly and second, by exhibiting low leaf laminar conductance behaving as a capacitor, a high KL and low gs are ensuring and extremely conservative stomatal behavior, opening a steady and reliable water supply to transpiring leaves. stomata only when soil moisture was plentiful (after rains) or when transpirational demand was minimal (early morn- ings, occasionally at night). Since most investigated Pe- Climate of Extant Pereskia Species reskia have been shown to exhibit some degree of CAM Not all of the Pereskia specimens that we collated from cycling, it is probably this ability that enables such con- various sources had enough location information to ac- servative stomatal behavior. Recycling respired carbon curately determine latitude/longitude coordinates; we were when stomata remain closed during the day allows the confident in using a total of 330 data points. Numbers of plant to maintain basic metabolic processes without re- locations for each species varied, ranging fromn p 1 in lying heavily on carbon stores (Kluge and Ting 1978; Ting Pereskia quisqueyana and P. marcanoi (which are endan- 1985; Cushman 2001). There is also some evidence of gered and currently known only from single populations) inducible CAM photosynthesis, where stomata open at ton p 73 in . Most climate variables are night, in three of the seven Pereskia species examined in similar for extant Pereskia species; figure 4 shows the spe- this study (P. guamacho, P. diaz-romeroana, and P. sa- cies ranges of mean annual precipitation and mean annual charosa). The distribution of inducible traits is not straight- temperature. The environmental parameters of most spe- forward to reconstruct, however, because of the possibility cies correspond to the tropical dry and very dry forest life of “false negatives”; just because the trait was not induced zones of Holdridge (1967). The only exception to this is in the course of a given study does not mean that the trait Pereskia bleo, which lives in areas receiving considerably is never induced. Since CAM cycling has been recorded higher annual rainfall. All species live in climates with previously in seven of 11 investigated Pereskia species, and highly seasonal precipitation patterns, with the mean wet- here we present some evidence of inducible full CAM in test month receiving 187 mm and the driest month 26 Pereskia species from both major Pereskia clades, we con- mm. clude that the CAM photosynthetic pathway was also pre- sent in ancestral Pereskia and played a limited but poten- tially important role in allowing for conservative stomatal Ancestral Trait Reconstructions behavior and possibly in extending leaf life span. Results from our reconstruction analyses are presented in tables 2 and 3. Estimated values were surprisingly robust Discussion to variations in both branch lengths and the model of evolution used in reconstruction (range of values reported As our knowledge of phylogeny improves, taxa once in table 3; in table 2 we report only the mean value of all thought to be monophyletic are sometimes found to be analyses because of the small variation between them; re- basal paraphyletic assemblages (see Donoghue 2005). Such ported error estimates in table 3 are those associated with is the case for “Pereskia” in relation to the core cacti. running one analysis across all 500 trees with randomized Discoveries of this sort provide us with opportunities to branch lengths). This is probably because of the relatively dissect the sequence of evolutionary events through major Table 3: Estimated values for nodes A and B of environmental parameters, using different models of phenotypic evolution Maximum Minimum Precipitation of Precipitation of Precipitation temperature of temperature of Mean annual wettest month driest month seasonality index Mean annual Temperature warmest month coldest month Model of evolution precipitation (mm) (mm) (mm) (SD # 100) temperature (ЊC) seasonality index (ЊC) (ЊC) Node A: 00. ע 17.9 00. ע 32.9 12. ע 133 00. ע 25.6 09. ע 57.1 06. ע 27.8 05. ע 177 13. ע BM (a p .0) 1,256 00. ע 17.9 00. ע 32.9 09. ע 131 00. ע 25.7 07. ע 55.8 05. ע 28.6 03. ע 176 07. ע O-U (a p .5) 1,258 00. ע 17.9 00. ע 32.9 08. ע 132 00. ע 25.7 06. ע 57.7 04. ע 27.3 03. ע 177 07. ע O-U (a p 1.0) 1,256 00. ע 17.9 00. ע 32.9 09. ע 129 00. ע 25.7 06. ע 60.2 04. ע 25.5 03. ע 178 03. ע O-U (a p 5.0) 1,253 00. ע 18.0 00. ע 32.9 07. ע 127 00. ע 25.7 06. ע 61.0 05. ע 24.8 04. ע 179 02. ע O-U (a p 10.0) 1,252 Node B: 00. ע 12.2 00. ע 30.6 07. ע 175 00. ע 21.8 05. ע 69.2 08. ע 19.9 04. ע 174 08. ע BM (a p .0) 1,102 00. ע 12.2 00. ע 30.6 04. ע 174 00. ע 21.8 03. ע 70.3 05. ע 18.0 02. ע 173 03. ע O-U (a p .5) 1,100 00. ע 12.2 00. ע 30.6 05. ע 175 00. ע 21.8 03. ע 69.0 03. ע 19.7 02. ע 173 03. ע O-U (a p 1.0) 1,100 00. ע 12.3 00. ע 30.6 04. ע 175 00. ע 21.9 01. ע 68.1 01. ע 20.4 01. ע 173 03. ע O-U (a p 5.0) 1,099 00. ע 12.4 00. ע 30.5 03. ע 175 00. ע 21.9 01. ע 68.3 01. ע 20.2 00. ע 172 02. ע O-U (a p 10.0) 1,098 SE of analyses run over 500 trees with identical topologies but randomized branch lengths.BM p Brownian motion; O-U p Ornstein -Uhlenbeck;a p strength of ע Note: Values are presented asmean restraining force. 788 The American Naturalist

Figure 6: Ecological water use traits of ancestral Pereskia nodes and core cacti. evolutionary transitions and therefore to analyze how be difficult to infer for the node subtending the core cacti changes in form and function were related to one another. because there are no Cactoideae with functional leaves to Our analysis illustrates this approach and also highlights include in a broader sampling of these traits. Nevertheless, some of the difficulties encountered when trying to infer we may consider ancestral values of these leaf-related traits ancestral conditions with necessarily limited data and in at Pereskia nodes A and B with regard to how they influ- lineages that have undergone radical morphological shifts ence patterns of whole-plant water use, and we may com- during their evolution. pare these organismal-level (rather than organ-level) traits With our current knowledge of basal cactus phylogeny with what we know about the core cacti from the literature and character distributions in Pereskia and core cacti, our (fig. 6). Traits of this sort include habitat characteristics, abilities to confidently infer the ancestral conditions of tolerated tissue water deficits, stomatal behavior, and pho- particular traits fall along a spectrum: some inferences are tosynthetic water use efficiency. Addressing the problem straightforward, some are effectively impossible at this in this way, we are able to present a preliminary hypothesis stage, and others fall between these two extremes. For of the ecological niche and water relations of the first cacti. instance, based on the of Pereskia and the nested Resolution of major relationships within the Cactoideae position of Cactaceae within the , we can and Opuntioideae will provide a better basis for more now infer with considerable confidence that the first cacti targeted trait sampling of core cacti taxa to include (with were shrubs or small trees with photosynthetic leaves. At additional physiological data from Maihuenia and Por- the other end of the spectrum, the lack of functional leaves tulacaceae outgroups) in further tests of this hypothesis. in all extant Cactoideae and most Opuntioideae makes it The reconstructed ecology and water use strategies of unlikely that we will ever assess with great accuracy any ancestral nodes A and B are remarkably similar to those particular physiological parameter of how those leaves that typify leafless, stem-succulent cacti (fig. 6), implying functioned. Values of KL, HV, and SLA, for example, will that the cacti had been inhabiting their particular ecolog- Origin of the Cactus Life-Form 789 ical niche long before they evolved the anatomical spe- A Preliminary Sketch of Early Cactus Evolution cializations that are often associated with their survival in water-limiting environments. In other words, it may not In fact, we can now begin to piece together the order of have been the evolution of leaflessness and stem succulence the early events that culminated in the evolution of the in cacti that yielded the cactus ecological water use strategy; cactus life-form (fig. 7). We have argued that the first cacti instead, it seems that it was the origin of this strategy in lived in tropical, subhumid to semiarid environments and these ancestors that promoted the evolution of leaflessness exhibited the water use strategy described above. After the and stem succulence. first split within Cactaceae, the caulocactus lineage may We are not suggesting that there are no ecological or have moved into drier and cooler climates with greater physiological differences between Pereskia species and the seasonality of both precipitation and temperature. During core cacti. Certainly, the core cacti are superior to their this time, the caulo cacti evolved delayed bark formation leafy relatives both at storing water and in using it effi- and stem stomata, key traits in the early development of ciently. They also occupy a greater diversity of habitats, the stem as a photosynthetic organ. Data from extant Pe- with many species experiencing more extreme drought reskia suggest that these traits are not by themselves suf- than either extant Pereskia or their reconstructed ancestors. ficient for significant stem photosynthesis (Gibson and Indeed, our work implies that the core cacti are so suc- Nobel 1986; Nobel and Hartsock 1986; Martin and Wallace cessful in these environments precisely because of their 2000); it appears that modifications of the stem cortical anatomical specializations. That is, the water use strategy tissue were also necessary (for discussion, see Edwards et shared by Pereskia and the core cacti, which we are arguing al. 2005). Opuntioideae, Cactoideae, and Maihuenia all was also present in the first Cactaceae, would be much exhibit significantly greater water storage in leaf (leaves improved with a longer-lived photosynthetic tissue system are present in some Opuntioideae and in Maihuenia,a (provided by stem-based, rather than leaf-based, photo- specialized, leafy of cold, arid ), synthesis) and larger reservoirs of stored water (allowing stem, and/or root tissues than any Pereskia species (Gibson stomata to open for longer periods, in stronger drought and Nobel 1986; Leuenberger 1997; Mauseth 1999), in- situations). dicating a shift toward increased succulence before the divergence of the two major core cactus clades. It is within Our results are also consistent with the view that eco- the Cactoideae that stem succulence is most fully devel- logical traits can be more highly conserved in evolution oped, and it is only in this lineage that the stem cortex than generally assumed (Webb et al. 2002; Wiens and Don- has evolved vascular bundles. It has been suggested (Mau- oghue 2004). Studies of adaptive radiations focus our at- seth and Sajeva 1992) that the vascularization of the cortex tention on just how evolutionarily labile ecological traits may have been a “key innovation” associated with the can be under some circumstance (e.g., in lineages that have increased diversity and disparity of the Cactoideae, which colonized relatively “open” islands). In contrast, some re- contains the vast majority of cactus species (≈1,250/1,600 cent analyses have drawn attention to cases in which in- species) and exhibits the greatest morphological diversity. terspecific competition and ecological sorting have pre- This remains to be tested, and the precise causal link be- dominated over in situ evolutionary adaptation in the tween this feature and diversity/disparity is currently un- assembly of ecological communities (Peterson et al. 1999; clear to us. In any case, however, distributing vascular Webb 2000; Ackerly 2004; Feild et al. 2004; Davis et al. tissue throughout the inner and outer cortex allows for a 2005; but see Cavender-Bares et al. 2004). In the case of more efficient connection between photosynthetic cells Cactaceae, we argue that the “cactus ecological niche,” at (outer cortex) and water and photosynthate storage cells least with respect to water use, was occupied for some time (inner cortex, wood, and pith). This results in more rapid before the evolution of the morphological and anatomical translocation of both water and photosynthetic assimilates specializations that are often assumed to be responsible throughout the cactus stem than could occur with cell-to- for their water use strategy. This niche conservatism may cell diffusion, and it may relax possible transport con- even predate the divergence of the cactus lineage from its straints on the ultimate width, shape, and water storage portulacaceous ancestors because most Portulacaceae in- capacity of the cactus stem. habit warm, water-limiting environments, are often suc- Reconstructing the evolution of perhaps the two most culent, and exhibit varying degrees of CAM photosynthesis characteristic features of the cactus life-form—stem-based (Martin and Zee 1983; Carolin 1987, 1993; Martin and photosynthesis and a functional loss of leaves—is more Harris 1993; Eggli and Ford-Werntz 2001; Guralnick and complicated. Given the apparent lack of stem-based pho- Jackson 2001). When placed in this broader phylogenetic tosynthesis in Maihuenia (Martin and Wallace 2000), it is context, the evolution of leaflessness and stem succulence possible that this trait arose independently in Opuntioi- does not seem quite so extraordinary. deae and Cactoideae. It is more certain that the functional 790 The American Naturalist

Figure 7: Overview of events in early cactus evolution. The placements of traits listed in white boxes are more certain; the placements of traits in gray boxes are more speculative. The gray line tracing several internal branches indicates that the traits in the gray box to the right could have evolved anywhere along those branches. Multispecies lineages are represented by triangles that are scaled to roughly represent relative lineage diversity (e.g., Cactoideae includes approximately 80% of all cactus species). loss of leaves evolved independently in the two lineages, Opuntioideae, then, that may hold the key to understand- as evidenced by a number of species of Opuntioideae that ing the processes involved in completely transferring the produce functional leaves (along with photosynthetic photosynthetic function from the leaves to the stem. How- stems) and are presumed to represent early-branching ever, we currently lack adequate phylogenetic resolution opuntioid lineages (Wallace and Dickie 2002). It is the of the major opuntioid lineages, and we know little about Origin of the Cactus Life-Form 791

field water relations of leafy opuntioid taxa such as Pe- Bucci, S. J., G. Goldstein, F. C. Meinzer, F. G. Scholz, A. C. Franco, reskiopsis and . Just as combining phylogenetic and M. Bustamante. 2004. Functional convergence in hydraulic and physiological analyses of Pereskia has provided insights architecture and water relations of tropical savanna trees: from leaf to whole plant. Tree Physiology 24:891–899. into early cactus evolution, we expect that such studies Butler, M. A., and A. A. King. 2004. Phylogenetic comparative anal- within the Opuntioideae will be especially helpful in un- ysis: a modeling approach for adaptive evolution. American Nat- derstanding these “final steps” in the evolution of the cac- uralist 164:683–695. tus life-form. Butterworth, C. A., and R. S. Wallace. 2005. Molecular of the leafy cactus Pereskia (Cactaceae). Systematic Acknowledgments 30:800–808. Carolin, R. 1987. A review of the family Portulacaceae. Australian We wish to thank Centro de Investigaciones en Ecologı´a Journal of Botany 35:383–412. de Zonas A´ ridas (Coro, Venezuela), the Jardin Botanico ———. 1993. Portulacaceae. Pages 544–555 in K. Kubitzki, ed. Fam- Nacional (Santo Domingo, ), the Her- ilies and genera of vascular plants. Springer, Heidelberg. Cavender-Bares, J., D. D. Ackerly, D. A. Baum, and F. A. Bazzaz. bario Nacional de Bolivia (HNB; La Paz, Bolivia) and the 2004. Phylogenetic overdispersion in Floridian oak communities. Museo de Historia Natural Noel Kempff Mercado (Santa American Naturalist 163:823–843. Cruz, Bolivia) for their kind hospitality and logistical sup- Cunningham, C. W., K. E. Omland, and T. H. Oakley. 1998. Recon- port. Friar Andres and Father Gee, S. Beck, M. Diaz, C. structing ancestral character states: a critical reappraisal. Trends in M. Dunn, J. Edwards, E. Guzman and the Guzman family, Ecology & Evolution 13:361–366. P. Lundholm, R. I. Meneses and Gisela of HNB, and Z. Cushman, J. C. 2001. Crassulacean acid metabolism: a plastic pho- Quintana all helped tremendously with data collection in tosynthetic adaptation to arid environments. 127: 1439–1448. the field. M. Lerdau deserves a special thank-you for his Davis, C. C., C. O. Webb, K. J. Wurdack, C. A. Jaramillo, and M. J. generous long-term loan of a Li-Cor 1600 steady state Donoghue. 2005. Explosive radiation of supports a porometer. We are also grateful to N. Cellinese, C. W. mid- origin of modern tropical forests. American Dunn, N. M. Holbrook, and all members of the Donoghue Naturalist 165:E36–E65. lab for various discussions that have significantly influ- Diaz, M. 1984. Estudios fisoecologicos de 4 especies de cactaceas en enced this article. Two anonymous reviewers provided condiciones naturales. Master’s thesis. Instituto Venezolano de In- comments that greatly improved our manuscript. This vestigaciones Cientificas, Caracas. work was funded in part by a Deland Award for Student Diaz, M., and E. Medina. 1984. Actividad CAM de cactaceaes en con- diciones naturales. Pages 98–113 in E. Medina, ed. Eco-fisiologia de Research (Arnold , Harvard University) and a plantas CAM. Centro Internacional de Ecologia Tropical, Caracas. National Science Foundation Graduate Research Fellow- Donoghue, M. J. 2005. Key innovations, convergence, and success: ship. macroevolutionary lessons from plant phylogeny. Paleobiology 31: 77–93. Literature Cited Donoghue, M. J., and D. D. Ackerly. 1996. Phylogenetic uncer- tainties and sensitivity analyses in comparative biology. Philo- Ackerly, D. D. 2004. Adaptation, niche conservatism, and conver- sophical Transactions of the Royal Society of London B 351: gence: comparative studies of leaf evolution in the California chap- 1241–1249. arral. American Naturalist 163:654–671. Eamus, D., and L. Prior. 2001. Ecophysiology of trees of seasonally Ackerly, D. D., and P. B. Reich. 1999. Convergence and correlations dry tropics: comparisons among phenologies. Advances in Eco- among leaf size and function in plants: a comparative test logical Research 32:113–197. using independent contrasts. American Journal of Botany 86:1272– Edwards, E. J., and M. Diaz. 2006. Ecological physiology of Pereskia 1281. guamacho, a cactus with leaves. and Environment 29: Anderson, E. F. 2001. The cactus family. Timber, Portland, OR. 247–256. Barcikowski, W., and P. S. Nobel. 1984. Water relations of cacti during Edwards, E. J., R. Nyffeler, and M. J. Donoghue. 2005. Basal cactus desiccation: distribution of water in tissues. Botanical Gazette 145: phylogeny: implications of Pereskia paraphyly for the transition 110–115. to the cactus life form. American Journal of Botany 92:1177– Brodribb, T. J., and T. S. Feild. 2000. Stem hydraulic supply is linked 1188. to leaf photosynthetic capacity: evidence from New Caledonian Eggli, U., and D. Ford-Werntz. 2001. Portulacaceae. Pages 370–432 and Tasmanian rainforests. Plant Cell and Environment 23:1381– in U. Eggli, ed. Illustrated handbook of succulent plants: dicoty- 1388. ledons. Springer, Berlin. Brodribb, T. J., N. M. Holbrook, and M. V. Gutierrez. 2002. Hydraulic Farquhar, G. D., M. H. O’Leary, and J. A. . 1982. On the re- and photosynthetic co-ordination in seasonally dry tropical forest lationship between carbon isotope discrimination and the inter- trees. Plant Cell and Environment 25:1435–1444. cellular concentration in leaves. Australian Journal Brodribb, T. J., N. M. Holbrook, E. J. Edwards, and M. V. Gutierrez. of Plant Physiology 11:539–552. 2003. Relations between stomatal closure, leaf turgor and xylem Feild, T. S., N. C. Arens, J. A. Doyle, T. E. Dawson, and M. J. Don- vulnerability in eight tropical dry forest trees. Plant Cell and En- oghue. 2004. Dark and disturbed: a new image of early angiosperm vironment 26:443–450. ecology. Paleobiology 30:82–107. 792 The American Naturalist

Felsenstein, J. 1985. Phylogenies and the comparative method. Amer- Meinzer, F. C. 2003. Functional convergence in plant responses to ican Naturalist 125:1–15. the environment. Oecologia (Berlin) 134:1–11. Gibson, A. C., and P. S. Nobel. 1986. The cactus primer. Harvard Mooney, H. A., S. A. Bullock, and J. R. Ehleringer. 1989. Carbon University Press, Cambridge, MA. isotope ratios of plants of a tropical dry forest in . Func- Graham, C. H., S. R. Ron, J. C. Santos, C. J. Schneider, and C. Moritz. tional Ecology 3:137–142. 2004. Integrating phylogenetics and environmental niche models Nobel, P. S. 1977. Water relations and photosynthesis of a barrel to explore speciation mechanisms in dendrobatid frogs. Evolution cactus, acanthodes, in the Colorado desert. Oecologia 58:1781–1793. (Berlin) 27:117–133. Guralnick, L. J., and M. D. Jackson. 2001. The occurrence and phy- ———. 1988. Environmental biology of and cacti. Cambridge logenetics of crassulacean acid metabolism in the Portulacaceae. University Press, New York. International Journal of Plant Sciences 162:257–262. Nobel, P. S., and T. L. Hartsock. 1986. Leaf and stem CO2 uptake in Hansen, T. F. 1997. Stabilizing selection and the comparative analysis the three subfamilies of the Cactaceae. Plant Physiology 80:913– of adaptation. Evolution 51:1341–1351. 917. Hardy, C. R., and H. P. Linder. 2005. Intraspecific variability and ———. 1987. Drought-induced shifts in daily CO2 uptake patterns timing in ancestral ecology reconstruction: a test case from the for leafy cacti. Physiologia Plantarum 70:114–118. Cape flora. Systematic Biology 54:299–316. Nyffeler, R. 2002. Phylogenetic relationships in the cactus family Holdridge, L. R. 1967. Life zone ecology. Rev. ed. Tropical Science (Cactaceae) based on evidence from trnK/matK and trnL-trnF se- Center, San Jose, Costa Rica. quences. American Journal of Botany 89:312–326. Peterson, A. T., J. Soberon, and V. Sanchez-Cordero. 1999. Conser- Kluge, M., and I. P. Ting. 1978. Crassulacean acid metabolism: anal- vatism of ecological niches in evolutionary time. Science 285:1265– ysis of an ecological adaptation. Springer, Berlin. 1267. Lerdau, M. T., N. M. Holbrook, H. A. Mooney, P. M. Rich, and Rayder, L., and I. P. Ting. 1981. Carbon metabolism in two species J. L. Whitbeck. 1992. Seasonal patterns of acid fluctuations and of Pereskia (Cactaceae). Plant Physiology 68:139–142. resource storage in the arborescent cactus excelsa in Reich, P. B., M. B. Walters, and D. S. Ellsworth. 1992. Leaf life-span relation to light availability and size. Oecologia (Berlin) 92:166– in relation to leaf, plant, and stand characteristics among diverse 171. ecosystems. Ecological Monographs 62:365–392. Leuenberger, B. E. 1986. Pereskia (Cactaceae). Memoirs of the New Sack, L., and M. T. Tyree. 2005. Leaf hydraulics and its implications York Botanical Garden 41:1–141. in plant structure and function. Pages 93–114 in N. M. Holbrook ———. 1997. Maihuenia: monograph of a Patagonian genus of Cac- and M. A. Zwieniecki, eds. Vascular transport in plants. Elsevier, taceae. Botanische Jahrbucher 119:1–92. Oxford. Luttge, U., E. Medina, W. J. Cram, H. S. J. Lee, M. Popp, and Sack, L., P. D. Cowan, N. Jaikumar, and N. M. Holbrook. 2003. The J. A. C. Smith. 1989. Ecophysiology of xerophytic and halophytic “hydrology” of leaves: co-ordination of structure and function in vegetation of a coastal alluvial plain in northern Venezuela. II. temperate woody species. Plant Cell and Environment 26:1343– Cactaceae. New Phytologist 111:245–251. 1356. Martin, C. E., and F. S. Harris. 1993. Nocturnal respiration rates and Santiago, L. S., G. Goldstein, F. C. Meinzer, J. B. Fisher, K. Machado, accumulation in five species of (Portulacaceae) D. Woodruff, and T. Jones. 2004. Leaf photosynthetic traits scale during CAM-cycling. Journal of Plant Physiology 141:762–764. with hydraulic conductivity and wood density in Panamanian for- Martin, C. E., and R. S. Wallace. 2000. Photosynthetic pathway var- est canopy trees. Oecologia (Berlin) 140:543–550. iation in leafy members of two subfamilies of the Cactaceae. In- Schluter, D., T. Price, A. O. Mooers, and D. Ludwig. 1997. Likelihood ternational Journal of Plant Sciences 161:639–650. of ancestor states in adaptive radiation. Evolution 51:1699–1711. Martin, C. E., and A. K. Zee. 1983. C3 photosynthesis and crassu- Smith, S. D. , R. K. Monson, and J. E. Anderson. 1997. Physiological lacean acid metabolism in a Kansas rock outcrop succulent, Tali- ecology of North American desert plants. Springer, Berlin. num calycinum Engelm. (Portulacaceae). Plant Physiology 73:718– Sobrado, M. A. 1986. Aspects of tissue water relations and seasonal 723. changes of leaf water potential components of evergreen and de- Martins, E. P. 2004. COMPARE: computer programs for the statistical ciduous species coexisting in tropical dry forests. Oecologia (Ber- analysis of comparative data. Ver. 4.6. Distributed by the author lin) 68:413–416. at http://compare.bio.indiana.edu/. Department of Biology, Indi- Sperry, J. S., V. Stiller, and U. G. Hacke. 2003. Xylem hydraulics and ana University, Bloomington. the soil-plant-atmosphere continuum: opportunities and unre- Martins, E. P., and T. F. Hansen. 1997. Phylogenies and the com- solved issues. Journal 95:1362–1370. parative method: a general approach to incorporating phylogenetic Ting, I. P. 1985. Crassulacean acid metabolism. Annual Review of information into the analysis of interspecific data. American Nat- Plant Physiology 36:595–622. uralist 149:646–667. Vendramini, F., S. Diaz, D. E. Gurvich, P. J. Wilson, K. Thompson, Mauseth, J. D. 1999. Anatomical adaptations to xeric conditions in and J. G. Hodgson. 2002. Leaf traits as indicators of resource-use Maihuenia (Cactaceae), a relictual, leaf-bearing cactus. Journal of strategy in floras with succulent species. New Phytologist 154:147– Plant Research 112:307–315. 157. Mauseth, J. D., and J. V. Landrum. 1997. Relictual vegetative ana- Wallace, R. S. 1995. Molecular systematic study of the Cactaceae: tomical characters in Cactaceae: the genus Pereskia. Journal of using DNA variation to elucidate cactus phylogeny. Plant Research 110:55–64. Bradleya 13:1–12. Mauseth, J. D., and M. Sajeva. 1992. Cortical bundles in the persis- Wallace, R. S., and S. L. Dickie. 2002. Systematic implications of tent, photosynthetic stems of cacti. Annals of Botany 70:317–324. chloroplast DNA sequence variation in the Opuntioideae. Pages Origin of the Cactus Life-Form 793

9–24 in D. R. Hunt, ed. Studies in the Opuntioideae (Cactaceae). Wiens, J. J., and M. J. Donoghue. 2004. Historical biogeography, Research. Hunt, Sherborne, UK. ecology and species richness. Trends in Ecology & Evolution 19: Webb, C. O. 2000. Exploring the phylogenetic structure of ecological 639–644. communities: an example for rain forest trees. American Naturalist Winter, K., and J. A. M. Holtum. 2002. How closely do the d13C 156:145–155. values of crassulacean acid metabolism plants reflect the propor-

Webb, C. O., D. D. Ackerly, M. A. McPeek, and M. J. Donoghue. tion of CO2 fixed during day and night? Plant Physiology 129: 2002. Phylogenies and community ecology. Annual Review of 1843–1851. Ecology and Systematics 33:475–505. Westoby, M., D. S. Falster, A. T. Moles, P. A. Vesk, and I. J. Wright. 2002. Plant ecological strategies: some leading dimensions of var- iation between species. Annual Review of Ecology and Systematics Associate Editor: Susanne S. Renner 33:125–159. Editor: Jonathan B. Losos